The field of the present invention encompasses molecular biology and membrane technology. Specifically, the present invention relates to artificial membrane scaffold proteins (MSPs) and methods of using the membrane scaffold proteins to stabilize, disperse and solubilize fully or partially hydrophobic proteins such as tethered, embedded or integral membrane proteins while maintaining the biological activities of those membrane proteins or to stabilize, disperse and solubilize proteins directly from membrane fragments or membranes into a mimic of the native membrane environment.
Several years ago we pursued structural and functional studies of lipids complexed with apolipoproteins and characterized these molecular assemblies by scanning probe microscopy, based on the adsorption of synthetic high density lipoprotein disks (rHDL, apo A-I) onto mica in an oriented manner (Carlson et al., 1997; Bayburt et al., 1998; Bayburt et al., 2000; Carlson et al., 2000). The diameters of the discoidal structures observed are approximately 10 nm with a height of 5.5 nanometers. The 5.5 nm high topology observed is most compatible with a single membrane bilayer epitaxially oriented on the atomically flat mica surface (Carlson et al., 1997).
Purified membrane proteins can be reconstituted into the phospholipid bilayer domain of certain such discoidal structures and studied in solution or subsequently adsorbed on a suitable surface for examination by structural or spectroscopic techniques that take advantage of a surface of oriented protein-bilayer assemblies. In the latter case, the underlying discoidal structures containing the membrane protein are easily recognizable and provide a point of reference for judging the quality of the sample and images.
High-density lipoproteins (HDL) are spherical assemblies of a protein component, termed apo A-I, and various phospholipids. HDL particles play an important role in mammalian cholesterol homeostasis by acting as the primary conduit for reverse cholesterol transport (Fielding and Fielding, 1991). The function of HDL as a cholesterol transporter relies upon the activity of the HDL-associated enzyme lecithin-cholesterol acyl transferase, or LCAT (Glomset, 1968; Jonas, 1991), which mediates the insertion of cholesterol esters into HDL lipoprotein particles. Certain portions of the apo A-I protein are required for the activity of this enzyme (Holvoet et al., 1995). In addition, a part of the apo A-I protein is thought to be in a globular domain at the N-terminus, and to be responsible for interactions with cell surface receptors. One nascent form of HDL particles has been assumed to be that of a discoid based on electron microscopy of stained preparations (Forte et al., 1971). Our laboratory has confirmed this using AFM studies of synthetic forms of rHDL under aqueous conditions. This form, however, is not the predominant form in circulation in vivo. Rather, the apo A-I structure appears to have evolved to stabilize a spherical form.
Two general models for the nascent structure of HDL disks have been proposed. One model has the apo A-I protein surrounding a circular bilayer section as a horizontal band or “belt” composed of a curving segmented alpha helical rod (Wlodawer et al., 1979). The other “picket fence” model has the protein traversing the edges of the bilayer vertically in a series of helical segments (Boguski et al., 1986). Both models are based primarily on indirect experimental evidence, and no three dimensional structure of the entire particle is available to distinguish between them.
The currently accepted model is the belt model, which is consistent with some electron microscopy and neutron scattering data (Wlodawer et al., 1979), where the helices are arranged longitudinally around the edge of the bilayer disks like a “belt” (Segrest et al. 1999). More recent infrared spectroscopy studies using a new method of sample orientation for dichroism measurements are more consistent with the belt model, in contrast to earlier studies (Wald et al., 1990; Koppaka et al., 1999). So far, there is no complete and direct evidence as to which model is correct, even though a low resolution x-ray crystal structure for apo A-I crystallized without lipid (Borhani et al., 1997) has been obtained. The low resolution crystal structure of an N-terminally truncated apo A-I shows a unit cell containing a tetrameric species composed of 4 helical rods which bend into a horseshoe shape and which combine to give a circular aggregate about 125×80×40 Å. It was suggested that a dimeric species in this belt conformation is capable of forming discoidal particles.
The information collected to date concerning the reverse cholesterol transport cycle and the maturation of HDL particles suggests that the apo A-I protein has unique properties that allow it to interact spontaneously with membranes resulting in the formation of lipoprotein particles. Initial apo A-I lipid binding events have been proposed (Rogers et al., 1998), but the mechanism for conversion of bilayer-associated forms to discoidal particles remains unclear. The available evidence suggests that the energy of stabilization of lipid-free apo A-I is fairly low and that there is an equilibrium between two conformers (Atkinson and Small, 1986; Rogers et al., 1998). One conformer may be a long helical rod, and the other may be a helical “hairpin” structure about half as long. It has been suggested that the low stabilization energy and conformational plasticity allow apo A-I to structurally reorganize when it encounters a lipid membrane, thus facilitating the structural changes that would have to take place in both the membrane and the protein to produce discreet lipoprotein particles (Rogers et al., 1998). Once these particles are formed, apo A-I may still undergo specific conformational changes that contribute to the dynamic functionality of the lipoprotein particles and interaction with enzymes and receptors. All of these interactions and changes take place at the protein-lipid interface and in specific topologies providing surface accessibility of critical residues. Thus, there is little reason to believe that apo A-I itself would be ideal for generating a stable, nanometer size phospholipid bilayer of controlled dimension. There is no prior evidence that any lipoprotein would have the desired property of the direct extraction and solubilization of membrane proteins directly from a crude membrane or membrane fragment preparation, as disclosed herein.
Different types of lipid aggregates are used for reconstitution and study of purified membrane proteins; these include membrane dispersions, detergent micelles and liposomes (FIG. 1). Purified systems for biochemical and physical study require stability, which is not always inherent in or is limiting in these systems. Liposomes are closed spherical bilayer shells containing an aqueous interior. Reconstitution into liposomes by detergent dialysis or other methods typically results in random orientation of the protein with respect to outer and lumenal spaces. Since ligands or protein targets are usually hydrophilic or charged, they cannot pass through the liposomal bilayer as depicted in FIG. 1, although this may be advantageous in some instances. Since both sides of the liposomal bilayer are not accessible to bulk solvent, coupling effects between domains on opposite sides of the bilayer are difficult to study. Liposomes are also prone to aggregation and fusion and are usually unstable for long periods or under certain physical manipulations, such as stopped flow or vigorous mixing. The size of liposomes obtained by extruding through defined cylindrical pore sizes polydisperse in size distribution rather than exhibiting a uniform diameter.
Liposomes also may present difficulties due to light scattering, and aggregation of membrane proteins present in the bilayer and thermodynamic instability (Angrand et al., 1997; Savelli et al., 2000). The surface area of a liposome is relatively large (105 to 108 Å2). To obtain liposomes with single membrane proteins incorporated requires a large lipid to protein molar ratio. Detergent micelles allow solubilization of membrane proteins by interaction with the membrane-embedded portion of the protein and are easy to use. Detergent micelles are dynamic and undergo structural fluctuations that promote subunit dissociation and often present difficulty in the ability to handle proteins in dilute solutions. An excess of detergent micelles, however, is necessary to maintain the protein in a non-aggregated and soluble state. Detergents can also be denaturing and often do not maintain the properties found in a phospholipid bilayer system. Specific phospholipid species are often necessary to support and modulate protein structure and function (Tocanne et al., 1994). Thus, the structure, function, and stability of detergent solubilized membrane proteins may be called into question. Since an excess of detergent micelles is needed, protein complexes can dissociate depending on protein concentration and the detergent protein ratio. By contrast, the MSP nanostructures of the present invention are more robust structurally, having a phospholipid bilayer mimetic domain of discrete size and composition and greater stability and smaller surface area than unilamellar liposomes. The disk structures allow access to both sides of the bilayer like detergents, and also provide a bilayer structure like that of liposomes.
There is a long felt need in the art for stable, defined compositions for the dispersion of membrane proteins and other hydrophobic or partially hydrophobic proteins, such that the native activities and properties of those proteins are preserved.